DataMiningProject

Problem Definition

• EEG Eye State identification is a kind of common classification problem.
• The finding from these studies are important and useful for human cognitive state classification.
• It can be use for:
  • Driving drowsiness detection
  • Epileptic seizure detection
  • Human eye blinking detection

EEG Eye State!

• The dataset consist of 14 EEG values and a value that indicate for the eye state.
• In usual case, the data describing EEG eye state belong to continuous type of time-series data.
• All data is from EEG measurement with the Emotiv EEG Neuroheadset.
• EEG is an observing system of electrophysiology which records the electrical movement of the brain.

Dataset

• There are 15 attributes. 14 are EEG values as shown as figure. And class label that is eyeDetection column.

• ‘1’ indicates the eye-closed and ‘0’ indicates the eye-open.

• Number of instances(rows) are 14980.

• Number of attributes(columns) are 15.

  

Implementation Details

Classification Models and Applied Methods

• Decision Tree using Gain Ratio: Holdout, bagging and boosting methods for Gain Ratio.
• Decision Tree using Gini Index: Holdout, bagging and boosting methods for Gini Index.
• Naive Bayes: Holdout, cross validation, bagging and boosting methods for Naive Bayes.
• Artificial Neural Network with 1 hidden layer: Holdout, bagging and boosting methods for ANN with 1 hidden layer.
• Artificial Neural Network with 2 hidden layer: Holdout, bagging and boosting methods for ANN with 2 hidden layer.
• Support Vector Machines: Holdout, bagging and boosting methods for Support Vector Machines.

Conclusion